bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

Insights into global planktonic diversity: Comparisons between phylogenetically meaningful units that account for time

Teofil Nakov, Jeremy M. Beaulieu, and Andrew J. Alverson Department of Biological Sciences University of Arkansas 1 University of Arkansas, SCEN 601 Fayetteville, AR 72701

Abstract

Metabarcoding has offered unprecedented insights into microbial diversity. In many studies, short DNA sequences are binned into consecutively higher Linnaean ranks, and ranked groups (e.g., genera) are the units of biodiver- sity analyses. These analyses assume that Linnaean ranks are biologically meaningful and that identically ranked groups are comparable. We used a meta-barcode dataset for marine planktonic to illustrate the limits of this approach. We found that the 20 most abundant marine planktonic diatom genera ranged in age from 4 to 134 million years, indicating the non- equivalence of genera because some had more time to diversify than others. Still, species richness was only weakly correlated with genus age, highlighting variation in rates of speciation and/or extinction. Taxonomic classifications often do not reflect phylogeny, so genus-level analyses can include phylogenet- ically nested genera, further confounding rank-based analyses. These results underscore the indispensable role of phylogeny in understanding patterns of microbial diversity. Keywords: diversification, metabarcoding, microbes, phylogeny

Preprint submitted to Bioarxiv July 24, 2017 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

1 With potentially millions of species occupying all the world’s aquatic and

2 terrestrial biomes, microbial species diversity is notoriously difficult to dis-

3 cover and catalog. Traditional approaches to species discovery are time and

4 labor intensive, and they miss species that cannot be cultivated in the lab [1].

5 The phylogenetic diversity of this undiscovered ”microbial dark matter” is

6 often characterized through community DNA sequencing of barcode genes.

7 A typical workflow includes DNA extraction from an environmental sam-

8 ple, PCR amplification of a barcode fragment, and high-throughput DNA

9 sequencing of the amplicon. Sequencing reads are clustered into operational

10 taxonomic units (OTUs) that are subsequently binned into consecutively

11 higher taxonomic ranks, and these ranked groups, in turn, are often the

12 focus of biodiversity assessments [2].

13 Linnaean names and ranks are often taken to mean more than what they

14 are: arbitrary taxon delimitations disconnected from evolutionary history.

15 The treatment of named ranks as anything other than arbitrary implies that

16 identically ranked groups are somehow comparable, encouraging comparisons

17 of their ecology, biogeography, and species richness [3, 4, 5]. The only mean-

18 ingful comparisons involve groups with comparable evolutionary histories [6].

19 In this sense, monophyletic groups (clades) are more likely to be biologically

20 cohesive units, and they should have comparable species richness if they are

21 similar in age and have diversified at similar rates [7]. Comparison of mono-

22 phyletic groups, while accounting for time, provides a robust framework for

23 detecting clades with exceptional species richness and comparing their func-

24 tional, ecological, or biogeographic breadth [8].

25 The Tara Oceans Project sequenced 18S-V9 metabarcode fragments from

2 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

26 plankton samples to characterize microbial communities and species richness

27 across the world’s oceans [9]. A total of 20 diatom genera accounted for nearly

28 99% of all diatom sequencing reads, and these genera were found to differ

29 in relative abundance, cell size, habitat preference, geographical distribution,

30 and species richness [2], however, it was unclear whether or not these patterns

31 deviated from expectations.

32 We focused our analyses on the genus-based patterns of species richness

33 and expected that older genera would be more species rich because they

34 have had a longer period of time to diversify [7]. We used a time-calibrated

35 phylogenetic tree of 1,151 diatoms [10] to calculate expected species richness

36 for the 20 most abundant marine diatom genera in the Tara Oceans survey

37 [2]. Given the crown age of diatoms [10], relative extinction (i.e., extinc-

38 tion/speciation) estimated from Cenozoic fossil species [11], and a minimum

39 approximation of total described and undescribed diatom diversity (30,000

40 species; [12]), we calculated a net diversification rate for diatoms (i.e., speci-

41 ation - extinction). We then used this rate to calculate the upper and lower

42 bounds of expected OTU richness for the 20 focal diatom genera [8].

43 The 20 diatom genera ranged in age from 4134 million years (My). OTU

44 richness was only weakly correlated with clade age (r=0.36, 95% CI=-0.1–

45 0.7, df=18, P=0.12), with 12 of the 20 genera falling within expectations for

46 OTU diversity given their age (Figure 1). The most abundant and OTU-rich

47 genus, Chaetoceros, was also the oldest (Figure 1a). The birthdeath diversi-

48 fication model predicted that Chaetoceros diversity should range between 47

49 and 6567 species—the Tara Oceans dataset recovered 644 OTUs, consistent

50 with expectations for a clade of this age (Figure 1b). Some of the most diverse

3 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

a b c Chaetoceros ● ● ● Chaetoceros Rhizosolenia ● ● ● Rhizosolenia Navicula ● * ●B F ● Navicula Leptocylindrus ● ● ● Leptocylindrus Coscinodiscus ● * ● ● Coscinodiscus Actinocyclus ● ● ● Actinocyclus Synedra ● * ●B F ● Synedra Thalassiosira ● * ● ● Thalassiosira Guinardia ● ● ● Guinardia Fragilariopsis ● ● ● Fragilariopsis Proboscia ● ● ● Proboscia Pseudonitzschia ● ● ● Pseudonitzschia Pleurosigma ● ●B ● Pleurosigma Corethron ● ● ● Corethron Planktoniella ● ● ● Planktoniella Eucampia ● ● ● Eucampia not ● ●B ● Haslea * monophyletic Haslea Attheya ● ● ● Attheya ● Polar Centric Minidiscus ● Radial Centric ● ● ● Minidiscus ● Raphid Pennate B Many benthic underdescribed ● ● ● Minutocellus ● Araphid Pennate F Many freshwater underestimated Minutocellus

150 120 90 60 30 0 1 5 10 25 50 100 250 1000 5000 1 5 10 25 50 250 1000 5000 Crown age (Ma bp) Tara swarms Algaebase taxa

Figure 1: Age and estimated taxon richness of the 20 most abundant marine planktonic diatom genera identified by the Tara Oceans metabarcode project [2]. Crown ages and uncertainty (grey bars) were estimated from 1000 bootstrap phylogenies [10] (a). Taxon richness was estimated from the number of OTU swarms in the Tara Oceans dataset (b) and the number of accepted species names in AlgaeBase [13] (c); lines delimit 95% confi- dence intervals of expected richness given the crown age of a clade, empirical extinction fraction, and diatom-wide estimate of the net diversification rate.

4 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

51 genera identified by metabarcoding (e.g., Corethron and Pseudo-nitzschia)

52 had OTU richness estimates that exceeded expectations. Assuming OTUs

53 correspond to species and that our estimates of clade age are not heavily bi-

54 ased, these genera have either exceptionally high speciation or low extinction

55 rates. Identifying the drivers of these patterns might offer new mechanistic

56 insights into phytoplankton diversification. Comparisons between OTU rich-

57 ness (Figure 1b) and accepted taxonomic names from AlgaeBase (Figure 1c)

58 showed expected discrepancies for lineages with substantial diversity in ben-

59 thic or freshwater habitats (e.g., Navicula; Figure 1b, B and F annotations;

60 Figure 1c, blue bars) and were also useful in highlighting clades that are

61 underdescribed at the species level (Figure 1c, green bars).

62 Metabarcoding identified Thalassiosira as one of the most abundant,

63 OTU-rich, and geographically widespread marine planktonic diatom gen-

64 era. A total of eight genera were detected in the Tara

65 Oceans project (, Lauderia, Minidiscus, Planktoniella, Porosira,

66 Shionodiscus, , and Thalassioisra), and these genera ranged in

67 age from 463 My (Figure 2). Thalassiosirales embodies many of the problems

68 with misappropriation of biological or evolutionary properties to taxa based

69 on their names [14]. The name Thalassiosira applies to a polyphyletic set

70 of species whose common ancestor dates to 63 My and gave rise to nearly

71 the full phylogenetic breadth Thalassiosirales diversity (Figure 2, diamond).

72 As a result, including Thalassiosira in genus-level analyses leads to highly

73 biased comparisons involving a genus that, in reality, corresponds roughly to

74 a taxonomic order (Figure 2). Moreover, four of the eight thalassiosiroid gen-

75 era detected by metabarcoding are nested within Thalassiosira, highlighting

5 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

Time (Ma bp) 120 100 80 60 40 20 0

Porosira pseudodenticulata Porosira glacialis Lauderia borealis Lauderia annulata Roundia cardiophora 9 Thalassiosira gessneri ● Conticribra guillardii Conticribra weissflogii ●8 Thalassiosira conferta Thalassiosira sp.ccmp1065 Cyclotella nana Cyclotella nana Cyclotella distinguenda Cyclotella stylorum Cyclotella striata Cyclotella atomus Cyclotella choctawhatcheeana Cyclotella sp.mc01 Cyclotella litoralis Cyclotella cf.scaldensis Cyclotella meneghiniana 7 Cyclotella cf.meneghiniana ● Cyclotella cryptica Cyclotella meneghiniana Cyclotella sp.wc032 Cyclotella sp.l042 Cyclotella gamma Thalassiosira hyalina Thalassiosira tenera Thalassiosira oceanica Skeletonema subsalsum Skeletonema potamos Skeletonema costatum Skeletonema grevillei Skeletonema ardens 6 ● Skeletonema menzellii Skeletonema tropicum Skeletonema pseudocostatum Thalassiosira minuscula Thalassiosira profunda Thalassiosira mediterranea Thalassiosira mala Thalassiosira curviseriata Thalassiosira nordenskioeldii Thalassiosira minima Thalassiosira angustelineata Thalassiosira lundiana Thalassiosira allenii Thalassiosira hendeyi Thalassiosira concaviuscula Thalassiosira baltica Thalassiosira angulata Bacterosira constricta Detonula confervacea 5 Bacterosira constricta ● Bacterosira bathyomphala Minidiscus trioculatus Minidiscus sp.ccl2009a ●4 Thalassiosira decipiens Thalassiosira pacifica Thalassiosira aestivalis Thalassiosira Thalassiosira sp.unc1203 Thalassiosira sp.ccmp1281 not Thalassiosira Detonula pumila ●3 Thalassiosira proschkinae Thalassiosira antarctica Thalassiosira sundarbana Thalassiosira delicata Thalassiosira rotula Thalassiosira delicatula MRCA: all Thalassiosira Thalassiosira punctigera Thalassiosira eccentrica MRCA: Clade with type Thalassiosira tumida Thalassiosira aff.antarctica of Thalassiosira Shionodiscus ritscheri Shionodiscus bioculatus 2 Shionodiscus oestrupii ● Thalassiosira nodulolineata Planktoniella sol 1 Marine genera Discostella pseudostelligera● Discostella woltereckii nested within Discostella sp.hk126 Discostella stelligera 'Thalassiosira': Discostella sp.hyk0210a2 Discostella sp.rg2012 Discostella pseudostelligera 1. Planktoniella Lindavia bodanica Lindavia costei 2. Shionodiscus Lindavia ocellata Lindavia cf.comensis Cyclostephanos tholiformis 3. Detonula Cyclostephanos invisitatus Cyclostephanos dubius 4. Minidiscus Stephanodiscus sp.rg2012 5. Bacterosira Stephanodiscus yellowstonensis Stephanodiscus reimerii 6. Skeletonema Stephanodiscus niagarae Stephanodiscus neoastraea 7. Cyclotella Stephanodiscus agassizensis Stephanodiscus parvus 8. Conticribra Stephanodiscus sp.fhtc11 Stephanodiscus minutulus 9. Roundia Stephanodiscus hantzschii Stephanodiscus binderanus

120 100 80 60 40 20 0 Time (Ma bp)

Figure 2: The genus Thalassiosira encompasses at least 10 marine planktonic diatom genera (including Thalassiosira) that range from6 4-63 My in age. Topology and divergence times are based on [10]. bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

76 a common source of non-independence in rank-based comparisons (Figure 2,

77 yellow branches). A more informative, phylogenetically based genus-level

78 classification may have revealed clade-specific habitat preferences or geo-

79 graphic distributions among the many distinct Thalassiosira lineages [15].

80 The problems with rank-based comparisons, including as they relate to di-

81 atoms, are well known [14, 16, 15]. A frequently cited advantage of metabar-

82 coding is that it does not require taxonomic expertise. Still, the taxonomic

83 affiliations of metabarcode sequences often become the units of biodiversity

84 analyses. A working knowledge of phylogeny and systematics—which invari-

85 ably highlight the deficiencies of Linnaean classifications—can lead to more

86 meaningful analyses that explicitly incorporate phylogenetic history, ensuring

87 robust comparisons of biologically equivalent units that account for time.

88 [1] C. Rinke, P. Schwientek, A. Sczyrba, N. N. Ivanova, I. J. Anderson, J.-F.

89 Cheng, A. Darling, S. A. Malfatti, B. K. Swan, E. A. Gies, et al., Insights

90 into the phylogeny and coding potential of microbial dark matter (2013).

91 [2] S. Malviya, E. Scalco, S. Audic, F. Vincent, A. Veluchamy, J. Poulain,

92 P. Wincker, D. Iudicone, C. de Vargas, L. Bittner, et al., Insights into

93 global diatom distribution and diversity in the worlds ocean, Proceed-

94 ings of the National Academy of Sciences 113 (2016) E1516–E1525.

95 [3] P. D. Cantino, K. de Queiroz, et al., Phylocode: a phylogenetic code of

96 biological nomenclature, 2000.

97 [4] F. Pleijel, G. W. Rouse, Ceci n’est pas une pipe: names, clades and

98 phylogenetic nomenclature, Journal of Zoological Systematics and Evo-

99 lutionary Research 41 (2003) 162–174.

7 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

100 [5] P. Sundberg, F. Pleijel, Phylogenetic classification and the definition of

101 taxon names, Zoologica scripta 23 (1994) 19–25.

102 [6] P. H. Harvey, M. D. Pagel, et al., The comparative method in evolu-

103 tionary biology, volume 239, Oxford University Press Oxford, 1991.

104 [7] T. Stadler, D. L. Rabosky, R. E. Ricklefs, F. Bokma, On age and species

105 richness of higher taxa, The American Naturalist 184 (2014) 447–455.

106 [8] S. Magall´on,M. J. Sanderson, Absolute diversification rates in an-

107 giosperm clades, Evolution 55 (2001) 1762–1780.

108 [9] S. Sunagawa, L. P. Coelho, S. Chaffron, J. R. Kultima, K. Labadie,

109 G. Salazar, B. Djahanschiri, G. Zeller, D. R. Mende, A. Alberti, et al.,

110 Structure and function of the global ocean microbiome, Science 348

111 (2015) 1261359.

112 [10] T. Nakov, J. M. Beaulieu, A. J. Alverson, A time-calibrated phylogeny

113 of diatoms (bacillariophyta), under review (2017).

114 [11] D. Lazarus, J. Barron, J. Renaudie, P. Diver, A. T¨urke, Cenozoic plank-

115 tonic marine diatom diversity and correlation to climate change, PLoS

116 One 9 (2014) e84857.

117 [12] D. G. Mann, P. Vanormelingen, An inordinate fondness? the num-

118 ber, distributions, and origins of diatom species, Journal of eukaryotic

119 microbiology 60 (2013) 414–420.

120 [13] M. Guiry, G. Guiry, Algaebase. national university of ireland, galway,

121 2017.

8 bioRxiv preprint doi: https://doi.org/10.1101/167809; this version posted July 24, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license.

122 [14] A. J. Alverson, B. Beszteri, M. L. Julius, E. C. Theriot, The model ma-

123 rine diatom likely descended from a freshwater

124 ancestor in the genus cyclotella, BMC evolutionary biology 11 (2011)

125 125.

126 [15] R. Wiese, J. Renaudie, D. B. Lazarus, Testing the accuracy of genus-

127 level data to predict species diversity in cenozoic marine diatoms, Ge-

128 ology 44 (2016) 1051–1054.

129 [16] J. P. Kociolek, and ecology: further considerations,

130 Proceedings-Califorina Academy of Sciences 56 (2005) 99.

9